Xplor Technologies powers the experiences at the heart of everyday life. Through modern vertical software, embedded payments, and AI-powered capabilities, we help businesses in fitness, recreation, golf and club, field services, laundry, education, and other membership-based and service-based industries simplify operations, uncover insights, and elevate customer and member experiences. 

We unite popular brands such as Clubessential, foreUP, myFitApp, Vermont Systems, Momence, Exerp, and many more.

Full-time · Remote · 3–5 years of experience

Why this role exists

This role is about engineers who build the AI features our users actually interact with — chat interfaces, agents, retrieval systems, AI-powered workflows inside our product. We're hiring a full stack engineer who can ship LLM-powered features end-to-end: from the prompt and the eval suite, through the retrieval and orchestration layer, to the UI a user clicks.

You should be the kind of person who has gone deep on LLMs, has strong opinions about evals, and rolls their eyes at demos that don't survive contact with real users.

If your strengths are general full stack engineering rather than LLM-specific work, see also our Full Stack Engineer role — it may be a closer fit.

If "AI feature" still means "ChatGPT wrapper" to you, this isn't your team. If you've already shipped something into production and watched it break in interesting ways, keep reading.

What you'll do

You'll own AI-powered features end-to-end — model choice, prompts, retrieval, tool use, the orchestration around it, the UI on top, and the evals that keep it from regressing. The work spans research-flavoured experimentation and hard product engineering, often in the same week.

Concretely, in your first six months you'd expect to:

• Ship at least one significant AI feature into production — agent, RAG flow, generative UI,

or similar

• Build out our eval and observability stack so we can ship LLM changes with confidence

rather than vibes

• Drive a measurable improvement on a key model-quality metric (accuracy, latency, cost,

hallucination rate)

• Help shape our internal AI engineering practices — model selection, prompt versioning,

regression testing

Day-to-day, you'll write code (a lot of it AI-assisted), design and evaluate prompts, debug agent

failures, partner with product and design on what AI features should even be, and make calls on

trade-offs between cost, latency, and quality.

What we're looking for

We're optimizing for engineers who are equally serious about software engineering and the new craft of building with LLMs. Plenty of people have one; we need both.

A genuine willingness to learn. The space moves weekly. The right model, framework, and pattern six months from now will not be the ones today. We expect you to keep up — not by chasing every release, but by knowing which signals matter.

Adaptability across stacks. No tribal identity around any one technology. Java, Node, Python, Go, TypeScript — whatever the problem calls for. "I haven't used that before" is a one-week problem, not a blocker.

Strong fundamentals. Data structures, algorithms, concurrency, networking, caching, transactions. Solid enough that you can apply them to unfamiliar problems instead of patternmatching on frameworks you've used before.

System design at both altitudes. You can sketch a high-level architecture for an AI-powered product — model boundaries, retrieval strategy, fallback paths, cost ceilings — and you can also do the low-level work: schema design, API contracts, prompt structure, hot-path optimization.

Sharp problem-solving. You break ambiguous problems into tractable pieces, you can hold a complex system in your head, and you debug from first principles. This matters double for AI work, where failure modes are subtle and "it usually works" is not a state you can ship from.

Hands-on experience building AI features in production. Not "I tried the OpenAI quick start." Real features that real users hit, with real consequences when they fail. Specifically, you've worked with several of:

• LLM APIs (Anthropic, OpenAI, or open-weight models via vLLM/Bedrock/Together) and have opinions on which to use when

• Retrieval-augmented generation: chunking strategies, embeddings, vector stores (pgvector, Pinecone, Weaviate, Qdrant), hybrid search

• Agentic systems: tool use, multi-step workflows, planning, MCP, frameworks like LangGraph/Inngest/Temporal or your own orchestration

• Structured output, function calling, and JSON-mode reliability

• Streaming responses and the UX patterns that make them feel good

• Prompt engineering as a serious discipline — versioning, A/B testing, regression suites, not vibes

Evals as a first-class skill. You know the difference between an eval that catches regressions and an eval that just makes you feel good. You've built golden datasets, designed LLM-as-judge pipelines (and know their failure modes), and used eval results to actually drive shipping decisions.

Production realism about LLM features. You think about latency budgets, cost per request, prompt caching, fallback chains, rate limits, content safety, prompt injection, and what to do when the model is wrong. You've debugged a hallucination at 2am at least once.

Strong full stack engineering chops. Real backend experience on an enterprise-grade framework (Spring Boot, .NET, NestJS, FastAPI, Django, or equivalent), and frontend experience in at least one of React, Vue, or Angular. You can build the UI for an AI feature, not just the API behind it. Streaming UIs, generative UI patterns, and chat interfaces are a plus.

Comfortable with both database paradigms. SQL (Postgres or MySQL) and NoSQL (DynamoDB, Mongo, or similar), plus a working knowledge of vector stores and when to reach for them.

Production cloud experience. Ideally AWS (Lambda, ECS, Step Functions, DynamoDB, S3, Bedrock, SageMaker), but the principles transfer. You can deploy and operate the things you build.

Fluency with AI-assisted development. Yes, also this. We expect you to be using Claude Code, Cursor, Copilot, or equivalent every day, and to have opinions on agentic loops vs. inline completions, scoping work for agents, reviewing AI-generated code critically, and building internal tooling around AI workflows. (If you're building AI features, you should also be expert at using AI to build them.)

High autonomy. Comfortable with ambiguous specs, async communication, and making calls

without a committee.

Nice to have

• Fine-tuning or post-training experience (LoRA, RLHF, DPO) — even small-scale

• Built or contributed to open-source AI infra, agent frameworks, or eval tooling

• Experience with model routing, prompt caching, or other cost/latency optimizations at scale

• Worked on AI safety, red-teaming, prompt injection defence, or content moderation pipelines

• Background in IR, NLP, or applied ML before the LLM era

• Strong writing — this team values it, and AI features live or die by precise language

• Prior experience in a small, fast-moving team (under ~30 engineers)

Got questions? You can email us at talentsupport@xplortechnologies.com.